PROPERTIES OF PORTLAND CEMENT. PART III. 235 



cement tends to reduce the plasticity. Otherwise, the differences between 

 the plasticity of non-seasoned and aerated cement pastes would be greater 

 than those shown by the preceding work. For the same reason, the set- 

 ting properties would have been affected to a greater extent. 



It also is known that the presence of lime in a state in which it is 

 immediately available as a soluble hydroxide, largely increases the 

 efficiency of calcium sulphate as a retarder. Therefore, as explained by 

 Candlot, the same substitution of calcium carbonate for calcium hydrate 

 tends again to increase the rate of set in plastered Portland cements. 



We have endeavored to show that two successive changes occur in the 

 condition of the free lime during the ordinary process of seasoning 

 Portland cement, the first of which tends to decrease, and the second to 

 increase, the normal rate of the set. In order to demonstrate the extent 

 to which these facts affect the problem under discussion, it is necessary 

 to take some additional facts into consideration. 



We have stated that the compounds which cause cements to set are 

 but slightly altered by exposiire to the atmosphere. This is true for the 

 time of exposure necessary to convert the available calcium oxide into 

 hydroxide. However, when Portland cements are undergoing thorough 

 aeration, the rate of set gradually decreases after no more available 

 slaked lime is present. Ultimately, after prolonged exposure, the cement 

 may finally become inert. In this respect coarsely ground, hard-burned 

 cements retain a normal set longer than finely ground material. 



Figure 20 shows the results obtained by thoroughly aerating a coarse 

 and a finely ground cement for 9 months. (See Table XXII, samples 1 

 and 2.) 



The coarsely ground cement gained its initial set in 7 hours while the finely 

 ground cement which had been seasoned in the same manner and during the same 

 time gained its initial set only after 23 days. 



Plate VI shows the same phenomenon in a specially interesting manner because 

 it illustrates a cause of unsoundness heretofore not mentioned in the literature 

 on the subject. 



Unsoundness in this instance was due to the fact that the finest particles in 

 the cement had become very inert, whereas the coarser ones still retained con- 

 siderable activity. When the pat was molded, the light, fine particles "floated" 

 to the upper surface and formed an outer covering about 1.5 millimeters thick 

 which failed to harden sufficiently to prevent the strains developed in the boiling 

 test from "blowing" and cracking it from the hardened cement underneath. 



Three changes in the rate of the set have been accounted for by the 

 tendency of water and carbon dioxide gradually to render the cement 

 itself inert. However, the work on commercial cements showed that in 

 some instances (see Table XXII, sample number 4) four changes take 

 place. When these occur the set is first accelerated, then retarded, again 

 accelerated and finally again retarded. The first set obtained is that of 

 a partially regauged cement paste. When a cement sets so quickly that 



